An Intellectual Brain MRI Image Retrieval System to aid the Diagnosis of Brain Tumors

Abstract

Human brain being a complex organ, detecting abnormalities like Brain Tumor, Alzheimer, and Schizophrenia etc. are not an easy task. A computer aided automated system called Content Based Medical Image Retrieval System (CBMIR) can be used to assist the medical practitioner in arriving at correct diagnosis. Brain tumor is a kind of disease related to the brain malfunction and goes through various stages. Brain tumor identification, classification in its initial stages is an important and challenging task. In this paper, focus is on revival of brain tumor images from large database where multiple stages of diseases are present. In order to realize this task, a feature extraction technique comprising of Local Binary Pattern (LBP), Gabor, and Histogram of Gradient (HOG) are used. Based on the attributes, Support Vector Machine (SVM) classifier is used for pattern learning and classification. An experiment is performed to measure the accuracy of SVM classifier and performance of the CBMIR system for different classes of brain tumor disease. With the extracted attributes, SVM achieves an accuracy of 89.33%, average precision of 89.35% and recall of 89.33%.